Problems with snd_pcm_status_get_delay()

Problems with snd_pcm_status_get_delay()

[James Courtier-Dutton]

Hello

Can someone please explain to me the use of snd_pcm_status_get_delay()

If a buffer size is 16384, and avail=6384, should delay therefore equil 
10000 ? In all cases ?

I have found that in  SND_PCM_STATE_RUNNING, delay = buffer_size - avail.
I have found that in  SND_PCM_STATE_PREPARED, delay = 0

This causes problems. As one fills the buffer in SND_PCM_STATE_PREPARED 
, delay stays at 0, but when the buffer fills enough and moves to state 
 SND_PCM_STATE_RUNNING, the delay value suddenly jumps to delay = 
buffer_size - avail.

This causes problems with applications that need "delay" to be as 
accurate as possible at all times.
Is this a bug in alsa09, or should I ignore "delay" and just use the 
"buffer_size - avail" value.

Obviously, in  SND_PCM_STATE_XRUN, delay is invalid.with avail values 
sometimes becoming > buffer_size!
so should not be used.

Cheers
James




-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: Problems with snd_pcm_status_get_delay()

[Takashi Iwai]

At Wed, 09 Oct 2002 02:25:40 +1000,
James Courtier-Dutton wrote:
> 
> Hello
> 
> Can someone please explain to me the use of snd_pcm_status_get_delay()
> 
> If a buffer size is 16384, and avail=6384, should delay therefore equil 
> 10000 ? In all cases ?

no, the current implementation is the state-sensitive, as you
mentioned below :

> I have found that in  SND_PCM_STATE_RUNNING, delay = buffer_size - avail.
> I have found that in  SND_PCM_STATE_PREPARED, delay = 0

delay=0 gets back at all states except RUNNING (and DRAINING on
playback).

> This causes problems. As one fills the buffer in SND_PCM_STATE_PREPARED 
> , delay stays at 0, but when the buffer fills enough and moves to state 
>  SND_PCM_STATE_RUNNING, the delay value suddenly jumps to delay = 
> buffer_size - avail.
> 
> This causes problems with applications that need "delay" to be as 
> accurate as possible at all times.
> Is this a bug in alsa09, or should I ignore "delay" and just use the 
> "buffer_size - avail" value.

snd_pcm_status_get_avail() returns the available space at any state.
so you can use this and calcuate "buffer_size - avail" manually
instead of get_delay().


ciao,

Takashi


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: Problems with snd_pcm_status_get_delay()

[Jaroslav Kysela]

On Tue, 8 Oct 2002, Takashi Iwai wrote:

> At Wed, 09 Oct 2002 02:25:40 +1000,
> James Courtier-Dutton wrote:
> > 
> > Hello
> > 
> > Can someone please explain to me the use of snd_pcm_status_get_delay()
> > 
> > If a buffer size is 16384, and avail=6384, should delay therefore equil 
> > 10000 ? In all cases ?
> 
> no, the current implementation is the state-sensitive, as you
> mentioned below :
> 
> > I have found that in  SND_PCM_STATE_RUNNING, delay = buffer_size - avail.
> > I have found that in  SND_PCM_STATE_PREPARED, delay = 0
> 
> delay=0 gets back at all states except RUNNING (and DRAINING on
> playback).

Delay functions should return error when the state is not correct. Fixed 
now.

						Jaroslav

-----
Jaroslav Kysela <perex@suse.cz>
Linux Kernel Sound Maintainer
ALSA Project  http://www.alsa-project.org
SuSE Linux    http://www.suse.com



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf